II see your reasoning to resave back to 24Bit after the dither so that in a side by side the file sizes would be the same and people wouldnt be able to just look at the file size to determine).
correct.
What would be a fair test would be to take the 24 Bit master and add gain ad then save 1 file at 24Bit and the other at 16 bit and do a blind test. But there are also some other serious considerations. Most folks will not have a playback system capable of really reproducing the 16Bits let alone 24.
If we add gain (say12db) to the 24bit master, the samples from 0 to -12db would be clipped/distorted; not a valid test. Adding a 16bit dither (bit reduction) to a 24bit file is the corect method and comparing the new file with the original 24bit is the correct method to compare the two.
If people are listening to shows through their crappy ass computer speakers than it really isnt going to be that dramatic of an improvement. If you move up to a higher class of playback it is very much like night and day. Bass response becomes much tighter and rounder in 24 Bits.).
I agree, crappy speakers, amps and a/d converter with poor signal to noise ratios make it impossible to hear the differences. The problem I have is I have very good speakers/headphones and soundcard and I am still guessing which is the 24bit vs. 16bit recording. I just disagree that it is night and day because most people, even on the best playback system, can\'t get it right consistently.
1 example that I can think of that really hammers the point home is this. On Pink Floyds Dark Side of the Moon, during Great big Gig in the Sky the female vocalist towards the end of the tune says "If you can hear me whispering your dying." On most average playback systems this is muddled and almost in-audible. On a HQ playback system it is clear as day.
you\'re just comparing good vs. bad playback system.
Same analogy holds true in 24Bit IMO. On the Beardslee castle recording I just made, Jordan is playing a solo or about to play a solo and he just ever so slightly tickles the keys at 1 point (very dramatic I thought for some weird reason) and again it is sooo crystal clear on the 24 Bit it is exactly where it should be in 3D. On the 16Bit it isnt as clear you really have to concentrate to hear it.
let me know what track #/ time and I will set up blind A/B test using a correct dither method.